Three-Dimensional Structure of a Halotolerant Algal Carbonic Anhydrase Predicts Halotolerance of a Mammalian Homolog

Three-dimensional structure of a halotolerant algal carbonic anhydrase predicts halotolerance of a mammalian homolog

Lakshmanane Premkumar*†, Harry M. Greenblatt†, Umesh K. Bageshwar*, Tatyana Savchenko*, Irena Gokhman*, Joel L. Sussman†‡, and Ada Zamir*‡

Departments of *Biological Chemistry and †Structural Biology, Weizmann Institute of Science, Rehovot 76100, Israel

Communicated by Ada Yonath, Weizmann Institute of Science, Rehovot, Israel, April 6, 2005 (received for review December 27, 2004) Protein molecular adaptation to drastically shifting salinities was broad range of salinities without exhibiting obligatory salt de- studied in dCA II, an ␣-type carbonic anhydrase (EC 4.2.1.1) from pendence (11–16). The structural principles discriminating the exceptionally salt-tolerant unicellular green alga Dunaliella halotolerant from mesophilic and halophilic proteins have yet to salina. The salt-inducible, extracellular dCA II is highly salt-tolerant be elucidated. A valuable source of such proteins is the unicel- and thus differs from its mesophilic homologs. The crystal structure lular green alga Dunaliella salina that proliferates in low to nearly of dCA II, determined at 1.86-Å resolution, is globally similar to saturating salt concentrations while attaining osmotic balance by other ␣-type carbonic anhydrases except for two extended ␣- intracellular accumulation of glycerol (17). Consequently, the helices and an added Na-binding loop. Its unusual electrostatic algal extracellular proteins should be able to cope with the properties include a uniformly negative surface electrostatic po- extremely broad spectrum of salinities sustaining algal growth. tential of lower magnitude than that observed in the highly acidic Two extracellular proteins identified in our studies that conform halophilic proteins and an exceptionally low positive potential at to this expectation are dCA I, a 60-kDa protein consisting of two (a site adjoining the catalytic Zn2؉ compared with mesophilic Ϸ52% identical, repeated ␣-type carbonic anhydrase (CA homologs. The halotolerant dCA II also differs from typical halo- domains (18), and the more recently discovered dCA II, an

philic proteins in retaining conformational stability and solubility Ϸ30-kDa, single-domain CA exhibiting Ϸ55% sequence identity BIOCHEMISTRY in low to high salt concentrations. The crucial role of electrostatic to each of the dCA I domains (19). features in dCA II halotolerance is strongly supported by the ability The CAs, found predominantly in animals but also in bacteria to predict the unanticipated halotolerance of the murine CA XIV and green algae, catalyze the reversible hydration of CO2 to form isozyme, which was conﬁrmed biochemically. A proposal for the bicarbonate and a proton. The catalytic mechanism employs a functional signiﬁcance of the halotolerance of CA XIV in the kidney Zn-bound hydroxide in the active site as a nucleophile in CO2 is presented. hydration to bicarbonate, a step followed by the regeneration of the Zn-bound hydroxide by intermolecular proton transfer via a nonhalophilic ͉ protein salt adaptation ͉ x-ray structure shuttling amino acid residue(s) from the Zn-bound water to a buffer molecule (20). An important corollary of this mechanism is the inhibition of these enzymes by various monovalent anions, ife in extreme or radically varying environments entails the Ϫ Ϫ molecular adaptation of cellular constituents, primarily pro- including Cl and Br , that displace the catalytically essential L ͞ teins, as safeguards against structural and functional damage. Zn-bound water hydroxyl and disrupt hydrogen-bonded net- Comparative structure–function studies of homologous proteins works at the active site, as indicated in crystal structures of of mesophilic, extremophilic, or extremotolerant origins unravel CA-anion adducts (21, 22). molecular strategies underlying the astounding acclimation The results presented here for dCA II provide insights into the power of proteins. Molecular adaptation of proteins to salt is of structural and electrostatic features that make a CA functionally particular interest considering the influence of salt on protein halotolerant. The lessons learned from the structure of the algal folding, oligomerization, solubility, and function (1). Most me- CA allowed us to predict and confirm the unexpected halotol- sophilic proteins function optimally at relatively low salinities erance of a membrane-associated mammalian homolog of vital and are adversely affected at high salinities. However, proteins importance in renal function. from extremely halophilic archaea, living in hypersaline envi- Materials and Methods ronments, typically require 1–2 M KCl or NaCl for proper folding and activity (1, 2). Halophilic proteins are characterized Enzyme Assays. Assays of affinity-purified dCA II (19), hCA I by a high negative surface electrostatic potential primarily due (Sigma), or mCA XIV (kindly provided by W. S. Sly, Saint Louis to the high abundance of surface acidic residues (1–6). Halo- University, St. Louis) for bicarbonate dehydration and esterase philic adaptation is best explained by the solvation–stabilization activities were performed as described in refs. 18 and 23. CO2 model whereby surface acidic residues bind hydrated ions to hydration activity was assayed at 22°C essentially as reported in refs. 18 and 23 with the following modifications: the reaction form a solvation shell that prevents water or salt enrichment at ⅐ the protein surface, thus allowing the proteins to remain soluble mixtures contained 25 mM Tris sulfate buffer (pH 8.0) and the and properly folded at high salt (1, 7, 8). These solvent–protein indicated concentrations of NaCl. The reaction was initiated by interactions are greatly dependent on the nature of the solvent

salts (7, 8). Complex salt bridge networks, weak protein–protein Freely available online through the PNAS open access option. interactions, and specific ion-binding were identified as addi- Abbreviations: CA, ␣-type carbonic anhydrase; CR, conserved region; VR, variable region; tional elements enhancing the stability and solubility of the VCR, variable conserved region. model halophilic protein, malate dehydrogenase from Haloar- Data deposition: The atomic coordinates and structure factors have been deposited in the cula marismortui (3, 7–10). Protein Data Bank, www.pdb.org (PDB ID code 1Y7W). A different response to salt is displayed by a versatile group of ‡To whom correspondence may be addressed. E-mail: [email protected] or halotolerant proteins, mainly from nonhalophilic as well as [email protected]. halophilic microbial sources that remain active throughout a © 2005 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0502829102 PNAS ͉ May 24, 2005 ͉ vol. 102 ͉ no. 21 ͉ 7493–7498 Downloaded by guest on October 1, 2021 the addition of 1.2 ml of CO2-saturated water and monitored Table 1. Effect of salt on the kinetic parameters of bicarbonate with a temperature-compensated pH meter interfaced to a dehydration activity of dCA II

personal computer. The initial rate was calculated from the Ϫ1 Ϫ1 [NaCI], M Km,* mM kcat,* (ms) kenz,(M⅐␮s) linear part of the pH change corrected by subtraction of the rate of the uncatalyzed reaction. dCA II 0 28.7 Ϯ 3.0 38.3 Ϯ 1.5 1.33 Structure Determination and Refinement. Crystals of recombinant 0.1 30.1 Ϯ 4.5 44.6 Ϯ 2.5 1.48 dCA II were obtained by using the vapor diffusion method (19). 0.25 32.2 Ϯ 7.9 41.2 Ϯ 4.0 1.27 The crystals were cryoprotected with paraffin oil. Native data 0.5 33.4 Ϯ 12.4 39.1 Ϯ 5.9 1.17 were collected in-house at 120 K by using R-AXIS IVϩϩ to 1.86-Å hCA I resolution. For phasing, multiwavelength anomalous dispersion 0 19.7 Ϯ 2.5 29.5 Ϯ 1.3 1.50 data sets were collected at 100 K at three wavelengths around the 0.1 80.1 Ϯ 11.4 28.7 Ϯ 2.3 0.36 zinc absorption edge on beamline BM14 at the European Steady-state kinetics parameters were determined essentially as described Synchrotron Radiation Facility. Diffraction data were processed in ref. 18. Assay mixtures contained the indicated concentrations of NaCI and with DENZO and SCALEPACK (24). Zinc positions were located on 0.28 ␮M hCA I or 0.54 ␮M dCA II. the basis of the anomalous difference. Experimental phases, to *Values are Ϯ SE. 2.0-Å resolution, were calculated from the multiwavelength anomalous dispersion data by using CNS (25), resulting in an overall figure of merit of 0.539. Phases were improved by centration, reaching an Ϸ2-fold increase at 2.0 M NaCl relative applying solvent-flipping density modification using CNS, result- to the activity in the absence of salt (Table 2). CO2 hydration ing in an overall figure of merit of 0.942. Based on the experi- assays indicated that the enzyme retained activity to at least 1.5 mental map, 92% of the initial model was built by using M NaCl (described below). Although measurements at higher XTALVIEW (26). A rigid body refinement in CNS using the initial salinities were constrained by the interference of salt in the assay, model and native amplitudes extended the resolution to 1.86 Å. dCA II activity was detectable even at 4.0 M NaCl. Similar Further refinement and model building was carried out in CNS responses to salt were observed in assays of dCA I (18). In and XTALVIEW, respectively. The data collection and refinement contrast, a representative mesophilic homolog, hCA I, was statistics are given in Table 5, which is published as supporting strongly inhibited already at 0.1 M NaCl. Specifically, at this salt information on the PNAS web site. concentration, the Km for bicarbonate increased Ϸ4-fold (Table 1), whereas the efficiency of esterase activity dropped by 75% Structural Analysis, Numerical Calculations, and Comparisons. The (Table 2) relative to the activity in the absence of salt. The following coordinate sets were obtained from the Protein Data exceptional halotolerance of dCA I and dCA II is further Bank (www.pdb.org): ID codes 1JCZ (human carbonic anhy- demonstrated in a comparison with eight additional mesophilic drase XII, hCA XII), 2CBA (human carbonic anhydrase I, hCA Ϫ CAs for Cl inhibition constants (see Table 6, which is published I), 1ZNC (human carbonic anhydrase IV, hCA IV), 1KOQ as supporting information on the PNAS web site). (Neisseria gonorrhoeae carbonic anhydrase, nCA), 1RAZ (human carbonic anhydrase II, hCA II), 1V9E (bovine carbonic anhydrase II, bCA II), 1RJ5 (murine carbonic anhydrase XIV, Overall Structure of dCA II. The x-ray structure of dCA II was mCA XIV). The contents of secondary structure elements were determined at 1.86-Å resolution (R factor, 16.7%; Rfree, 20.2%). calculated by DSSP (27). Surface accessibility was assessed by The asymmetric unit of the crystal contains two dCA II mole- DSSP using a per-residue cut-off of 25 Å2, with a probe radius of cules, Mol A and Mol B, related by noncrystallographic symmetry with a root-mean-square (rms) deviation of 0.25 Å for the 1.4 Å. Structure-based sequence alignments were carried out by ␣ using LSQMAN (28) and͞or SWISSPDBVIEWER (29). Determina- C atoms. The global fold of dCA II resembles that of other CAs tion of conserved regions (CRs), variable regions (VRs), and in its central antiparallel 10-stranded ␤-sheet, two ␣-helices, and ϩ variable conserved regions (VCRs) used a 2.7-Å cut-off crite- the presence of a catalytic Zn2 (Fig. 1). This similarity is rion. Molecular figures were generated in PYMOL (http:͞͞ exemplified by the structure-based sequence alignment of dCA pymol.org), and the surface electrostatic figures were generated II (274 aa) with the best-characterized human isozyme, hCA II in GRASP (30). The electrostatic potential calculations were (259 aa), which shows 24% sequence identity and a rms deviation ␣ performed by using DELPHI software (31). of 1.4 Å for 176 C atoms. Like other membrane-attached CAs Additional information related to materials and methods (34, 35), dCA II contains a single disulfide linkage between including details of electrostatic calculations can be found in Supporting Text, which is published as supporting information on the PNAS web site. Table 2. Effect of salt on the kinetic parameters of dCA II esterase activity

Results Ϫ1 [NaCI], M kenz,* (M⅐s) kenz, % of control Localization, Induction, and Enzymatic Activity of dCA II. The extracellular localization of dCA II in D. salina and its induction in dCA II response to increasing salinities or depletion of bicarbonate 0.0 66.2 Ϯ 1.0 100 resembled those reported in refs. 32 and 33 for dCA I. The 0.1 88.0 Ϯ 2.3 133 detailed results are presented in Supporting Text and also in Fig. 0.4 91.0 Ϯ 2.0 138 4, which is published as supporting information on the PNAS 1.0 102.0 Ϯ 1.0 155 web site. 2.0 121.0 Ϯ 2.1 183 The salt tolerance of dCA II was demonstrated in assays of hCA I three different CA activities. The steady-state kinetic measure- 0.0 498.0 Ϯ 7.5 100 ments of bicarbonate dehydration, determined at pH 6.8, indi- 0.1 120.0 Ϯ 2.1 24 cated that kenz and Km remained nearly constant in up to 0.5 M Enzymatic assays for esterase activity were performed essentially as de- NaCl (Table 1). Assays performed at acidic pH indicated that the scribed in ref. 18. Assay mixtures contained the indicated concentrations of IC50 for NaCl remained Ͼ0.5 M even at pH 5.6 (data not shown). NaCI and 0.54 ␮M hCA I or 3.37 ␮M dCA II. The kenz for esterase activity rose continuously with salt con- *Values are Ϯ SE.

7494 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0502829102 Premkumar et al. Downloaded by guest on October 1, 2021 Table 3. Comparison of amino acid distribution in solvent accessible surface area Polar, Nonpolar, Acidic, Basic, Acidic͞basic CAs % % % % ratio

dCA II 47.6 22.1 21.4 8.9 2.4 hCA XII 43.7 25.7 17.9 12.7 1.4 hCA II 34.5 19.3 21.9 24.2 0.9 nCA 39.6 21.0 14.9 24.6 0.6 hCA IV 33.4 22.4 18.0 26.2 0.7 hCA I 39.7 21.6 18.2 20.4 0.9 mCA V 33.4 24.6 22.1 19.8 1.1 bCA II 33.5 24.8 20.7 21.0 1.0

Calculations were performed as described in Materials and Methods. Polar: Asn, Gln, His, Ser, Thr, Tyr; nonpolar: Ala, Gly Trp, Pro, Phe, Leu, Ile, Val, Cys, Met; acidic: Asp, Glu; basic: Arg, Lys. The average solvent accessible surface areas for all the CAs is 11,710 Ϯ 766 Å2.

the buffers used in the purification and crystallization of the enzyme. To what extent L4 is specific for Naϩ or can bind other alkali cations remains to be established. To the best of our knowledge, dCA II is the only CA found to have a cation-binding loop. The noncatalytic additional Zn2ϩ is located at the interface of Fig. 1. Ribbon diagram of the dCA II structure. The regions corresponding to CRs (blue), VRs (green), and VCRs (red), as deﬁned in the text, are mapped onto the two noncrystallographic-symmetry-related molecules (Fig. the dCA II structure. Marked are the catalytic Zn2ϩ and insertions and deletions 5b and 6, which are published as supporting information on the in VCRs including L1 (the Zn binding loop), L4 (the Na-binding loop), and L5 as PNAS web site) and is tetrahedrally coordinated by two His-13 BIOCHEMISTRY ϩ well as the two extended ␣-helices (E and G). N and C termini are indicated. residues and two water molecules as shown in Fig. 5b. This Zn2 , and a hydrophobic patch of 567 Å2 at the dimer interface, suggested that dCA II assumes a dimeric form also in solution. Cys-31 and Cys-221 (equivalent to Cys-23 and Cys-201 in hCA However, analyses of an extensively dialyzed dCA II by gel- XII). filtration chromatography and atomic-absorption spectroscopy To identify structural features unique for dCA II, its structure were consistent with dCA II being a monomer containing a was compared with seven mammalian and one bacterial CA single Zn2ϩ (data not shown). structures deposited in the Protein Data Bank. In these eight structures, fully conserved structural elements, referred to as Surface Properties of dCA II Vis-a`-Vis Halophilic Proteins. The struc- ␣ CRs, were identified by pairwise structural alignments of C ture of dCA II was compared with other CA structures also in atoms, ignoring whether the sequences are similar or not. The variable regions that are mostly comprised of loops and turns remaining, structurally variable regions located mostly within localized at the protein surface. Table 3 shows that the solvent- surface loops are referred to as VRs. Although most CRs are accessible surface of dCA II differs from that of the other CAs conserved in the dCA II structure, some CRs, designated as in possessing a high ratio of acidic over basic amino acid residues VCRs, have undergone significant modifications and thus may resulting from a lowered level of basic residues, primarily lysines. be specific for dCA II. The mapping of VR, CR, and VCR At the same time, the level of acidic residues is kept similar to sections on the dCA II structure (Fig. 1) shows VCRs that differ that of the mesophilic homologs. The selective decrease of Lys from corresponding CRs by sequence deletions, insertions, and content in dCA II not only affects the surface charge but also amino acid replacements. The deletion in the VCR correspond- serves to decrease the surface hydrophobic character, as re- ing to L5 creates a kink at the position occupied by the ported in ref. 6 for halophilic proteins. proton-shuttling His-64 in highly efficient CAs, which is not The unusual surface amino acid composition of dCA II made present in dCA II. The insertion in the VCR corresponding to it of interest to compare its electrostatic surface potential with ␣E and the restructuring of a VR contiguous with ␣G extend that of other CAs (Fig. 2). The predominantly negative surface these two helices in dCA II with respect to the other CAs electrostatic potential of dCA II is strikingly different from the included in the comparison. Specifically, in dCA II ␣E contains uneven distribution of neutral, negative, and positive potentials 15 aa (Ala-166 to Asp-181), and ␣G is 14 aa (Arg-238 to characteristic of the other CAs, with the exception of the acidic Ala-251), as compared with hCA II where ␣E is 6 aa (Gly-156 ‘‘back’’ hemisphere of hCA XII. In this respect, dCA II surface to Val-161) and ␣G is 7 aa (Ser-220 to Phe-226) long. A resembles halophilic protein surfaces (1, 3–5, 7, 8). Nonetheless, quantitative comparison of secondary structure elements (see dCA II differs markedly from halophilic proteins primarily in its Table 7, which is published as supporting information on the ability to fold correctly (19) and to maintain solubility and PNAS web site) indicates that dCA II is significantly more enzymatic activity at low salt concentration (Tables 1 and 2). enriched in helical content, moderately poorer in ␤-strand Additionally, the far-UV CD and intrinsic fluorescence spectra structures, and similar in the proportion of loops and turns shown in Fig. 7, which is published as supporting information on relative to other CAs. the PNAS web site, indicate that the dCA II structure does not Other VCRs include the insertions that form loops L1 and L4 vary significantly in response to salinity shifts ranging from 0 to that bind a noncatalytic Zn2ϩ andaNaϩ, respectively (Fig. 1). 3.0 M NaCl, thus excluding salt-dependent conformational The identification of the bound Naϩ was based on the observed changes. octahedral coordination geometry and ligand distances (36) as In search of a plausible basis for the unusual salt responses of shown in Fig. 5a, which is published as supporting information dCA II, the density of charged amino acids on the protein surface on the PNAS web site. The bound Naϩ presumably originated in was compared with that of mesophilic homologs (Fig. 2) as well

Premkumar et al. PNAS ͉ May 24, 2005 ͉ vol. 102 ͉ no. 21 ͉ 7495 Downloaded by guest on October 1, 2021 Fig. 2. Electrostatic properties of dCA II and mesophilic CAs. The surface electrostatic potentials were calculated by using GRASP. The structures are graphically depicted looking down the active site cleft (ﬁrst row) or as the 180° rotated view (second row). The potentials are contoured to Ϫ2.5 kT per electron (red) and ϩ2.5 kT per electron (blue). The details of the electrostatic free energy (kcal͞mol) calculations are presented in Supporting Text. The densities of acidic and basic residues per 103Å2 were calculated from the ratio of the number of surface acidic (D and E) or basic (K and R) residues to total accessible surface area.

as with values reported for halophilic proteins (5). The halotol- of dCA II to anion inhibition, we calculated the electrostatic erant dCA II is outstanding among the CA structures compared potential at a site adjacent to the catalytic Zn2ϩ. This site in its low surface density of basic residues. However, the surface referred to as ‘‘Br-binding site’’ is defined by the position density of 2.33 acidic residues per 103Å2 is similar to that of the occupied by BrϪ in the structure of a Br–hCA II complex (22), mesophilic CAs with the possible exception of the rather basic studied to clarify the mechanism of halide inhibition of CA bacterial nCA. Conversely, 3 crystal structures and 185 homol- activity. The calculations (Table 4) reveal a markedly lower ogy models of halophilic proteins show an average density of 4.07 positive potential at the Br-binding site of dCA II relative to the acidic residues per 103Å2 as compared with an average of other CAs included in the comparison. The lower potential, most 2.50–2.86 acidic residues per 103Å2 for mesophilic proteins (5), probably arising from the negative surface potential of dCA II, a value close to that observed for dCA II. is likely to diminish the interaction efficiency of anions, e.g., ClϪ, The effect of salt on dCA II stability was assessed by solving with the active site, consequently enhancing salt tolerance. the nonlinear Poisson–Boltzmann equation for the entire protein molecule at various NaCl concentrations (Fig. 2). The loss of 9.3 Proposed Structural Basis for dCA II Salt Tolerance. A number of kcal͞mol in electrostatic free energy of dCA II on lowering the features of dCA II can be considered as potential determinants NaCl concentration from 3.0 to 0.1 M is within the range of salt tolerance. The predominantly negative surface electro- exhibited by the mesophilic CAs with which it was compared. A static potential endows the enzyme with the electrostatic prop- similar comparison of halophilic and mesophilic ferredoxins erties that, in analogy to halophilic proteins, can enhance transferred from 5.0 to 0.1 M NaCl indicated a free energy loss stability and solubility of the protein in high salt. Conversely, the of 17.3 kcal͞mol for the halophilic protein in contrast to a loss lower surface negative charge density of dCA II, close to that of of only 3.7 kcal͞mol calculated for its mesophilic homolog (37), mesophilic homologs, can enable the protein to fold correctly in keeping with the strong destabilization of halophilic proteins even at low salt. Long-range electrostatic interactions, probably at low salt concentrations. involving the surface negative charges, lower the positive po-

Active Site Structure. In addition to the need for maintaining the Table 4. Electrostatic potential at the Br-binding site global structure in varying salinities, dCA II should be exceptionally resistant to anion inhibition, because its mesophilic Potential at Br-binding site, counterparts are normally inhibited by low salt (Tables 1, 2, and CAs kT͞e 6). However, structural superposition of residues located within dCA II 11.0 Ϸ 2ϩ 8 Å from the catalytic Zn in dCA II and hCA II shows that hCA II 19.9 these residues are highly conserved and occupy essentially hCA I 20.0 identical positions (the detailed description is presented in nCA 20.0 Supporting Text; see also Fig. 8b, which is published as supporting hCA IV 17.5 2ϩ information on the PNAS web site). The catalytic Zn of dCA bCA II 23.3 II is coordinated with distorted trigonal bipyramidal geometry to hCA XII 24.5 three His residues, a water molecule, and the carboxylate oxygen of an acetate anion, originating in a buffer used during the The deﬁnition and coordinates of the Br-binding site are as described in the purification of the enzyme (Fig. 8a). Binding of acetate to the text. The coordinates for the various CAs were determined by superimposing the respective structures on the structure of hCA II complexed with Br (Protein active site, previously observed in the structures of hCA II and Data Bank ID Code IRAZ) (22). The approximate location of the Br-binding site hCA XII, is considered to mimic the binding of the bicarbonate corresponds to the Zn-bound water (Wat in Fig. 8a). The average potential [kT substrate (35, 38). per electron (e)] for 720 points within 0.2 Å around the Br-binding coordinate In another approach to understand the exceptional tolerance was calculated as described in Materials and Methods.

7496 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0502829102 Premkumar et al. Downloaded by guest on October 1, 2021 sponding activities of dCA II and hCA I, the results demonstrate the outstanding salt tolerance of mCA XIV, albeit to a slightly lower extent than that of dCA II or dCA I (18). This difference could be related to the absence in mCA XIV of elements such as the extended helices and the Na-binding loop as well as the slightly higher positive electrostatic potential compared with dCA II at the Br-binding site of mCA XIV. Discussion Similarly to the dCA I described in refs. 32 and 33, the salt- inducible single-domain dCA II is thought to act by converting bicarbonate, the major source of inorganic carbon available to Dunaliella,toCO2 that diffuses readily into the cells. As we have predicted (18), the enzymatic assays of dCA II conducted in low to multimolar NaCl concentrations indeed demonstrated its unusual salt tolerance. To the best of our knowledge, the dCA II crystal structure is the first reported 3D structure of a Dunaliella protein (www.ncbi.nlm.nih.gov͞Taxonomy) as well as of a bona fide halotolerant protein from any source. Spectral analyses of dCA II in solution indicated that it does not undergo meaningful conformational transitions as a function of salt concentration. Hence, the structure determined at low salt, as described here, should be valid for a broad range of salinities. Extensive studies of mammalian ␣-type CAs, emanating from their physiological and medical interest, included the determination of numerous 3D structures of free and inhibitor-

complexed enzymes. Thus, the halotolerant dCA II structure BIOCHEMISTRY Fig. 3. Electrostatic properties and effect of salt on enzymatic activity of could be critically viewed against the landscape of a wealth of mCA XIV. (a) The surface electrostatic potentials, the decrease in electrostatic structures of mesophilic homologs. Shared structural and func- free energy on transfer from 3.0 to 0.1 M NaCl, and surface density of acidic tional properties of halophilic proteins provided another frame and basic residues are as described in Fig. 2. (b and c) Assays of CO2 hydration of reference because no halophilic CA has been reported yet. activity (b) and ester hydrolysis activity (c) of dCA II (green), mCA XIV (yellow), or hCA I (red) were performed as described in Materials and Methods. Controls The basis proposed for the salt tolerance of dCA II considers indicate enzyme activity in the absence of salt. For esterase activity, 0.5 mM unique secondary structure elements likely to enhance its sta- p-nitrophenyl acetate was used as substrate. Pot, electrostatic potential. bility in high salt but primarily stresses the contribution of the negative electrostatic surface potential. The electrostatic interactions considered here could be partly akin to mechanisms tential at the Br-binding site, thus rendering dCA II less sensitive proposed for modulation of enzymatic activity by ‘‘steering’’ of to ClϪ inhibition. Furthermore, electrostatic repulsion by the substrates toward the active site (41–43) or modification of negative surface charges, particularly those close to the active interaction potentials within the active site (44). site entrance, might hinder the approach of inhibitory anions to The unanticipated discovery of the exceptional halotolerance the active site. In addition to its electrostatic properties, the two of mCA XIV upholds the predictions drawn from the dCA II extended surface ␣-helices and the Na-binding loop in dCA II structure and furthermore raises intriguing questions regarding probably serve as stabilizing factors in high salt. Extended helices the functional significance of the halotolerance of a mammalian have been implicated in the stability of thermophilic and halo- homolog. CA XIV has been identified in several tissues including philic proteins (4, 39). Specific bindings of ions such as Kϩ,Naϩ, brain, liver, and kidney. In the kidney, CA XIV has been or ClϪ have been implied in stabilizing intersubunit or crystal localized to the proximal convoluted tubule (40, 45, 46), partic- packing interfaces in several halophilic proteins (4, 9, 10). ularly in the apical plasma membrane, i.e., brush border (46), of Nonetheless, a Na-binding loop as found in dCA II has not been the S1 and S2 segments, where Ϸ90% of the bicarbonate in the reported yet in halophilic proteins. renal filtrate is reabsorbed into the blood. In this process, The critical role of the electrostatic properties in the salt protons secreted into the lumen, mainly by the brush border tolerance of the algal dCA II is strongly supported by the ability Naϩ͞Hϩ exchanger, protonate the luminal bicarbonate to car- of these properties to predict the unexpected halotolerance of a bonic acid that is dehydrated by a membrane-associated CA(s) phylogenetically remote homolog, the murine CA XIV. The to CO2 that easily permeates the brush border membranes. predominantly negative surface potential of the extracellular Within the cells, the reverse series of events, catalyzed by a catalytic domain of CA XIV (Fig. 3a), based on its recently cytoplasmic CA, generates bicarbonate that is reabsorbed into determined crystal structure (40), bears a strong resemblance to the blood across the basolateral membrane. Normally, the lumen that of dCA II (see Fig. 2). This initially observed similarity was pH falls from 7.4 at the glomerulus to Ϸ6.7 along the length of subsequently extended to several other characteristic properties the proximal tubule, as the bicarbonate is reabsorbed and the of dCA II, i.e., the low density of surface basic amino acid buffer capacity of the luminal fluid decreases. It is conceivable residues relative to acidic residues; the relatively small change in that the local pH at the brush border, the site of proton secretion, electrostatic free energy accompanying a transition from high to is even lower than that in the lumen itself, due to the restricted low salinity; and the lowered positive electrostatic potential, diffusion of protons or bicarbonate. albeit to a lower extent than in dCA II, at the Br-binding site (see The earlier view ascribing the membrane-associated luminal Figs. 2 and 3a). The power of these properties to predict the salt CA activity solely to CA IV has been revised upon the identi- tolerance of mCA XIV was unequivocally demonstrated in fication of CA XIV as a luminal CA localized mostly at distinct assays of CO2 hydration (Fig. 3b) and esterase (Fig. 3c) activities sites from CA IV (46). The question whether CA IV and CA in the presence of 0–3.0 M NaCl. Shown together with corre- XIV at the proximal convoluted tubule are functionally redun-

Premkumar et al. PNAS ͉ May 24, 2005 ͉ vol. 102 ͉ no. 21 ͉ 7497 Downloaded by guest on October 1, 2021 dant or physiologically distinct can now be considered in view of their salt tolerance while conserving the active site architecture the unexpected halotolerance of CA XIV. Because no osmotic and typical global fold except for modifications that potentially build-up is known to take place at the proximal convoluted increase the overall stability of the protein. Surface electrostatic tubule sections where the enzyme is localized, we considered the properties intermediate between those of halophilic and meso- possibility that acidification, by the mechanism proposed above, philic proteins play an essential role in the halotolerance of dCA was the critical environmental factor underlying the salt adap- II. Whether other halotolerant proteins share the surface elec- tation of CA XIV. As shown in our present and earlier studies, trostatic characteristics of dCA II must await further structure dCA II as well as dCA I remain almost resistant to anion Ϸ determinations. The large difference in anion sensitivity emerg- inhibition even at a pH of 5.5 (18), in contrast to the large ing from the disparate electrostatic properties of CA XIV and increase in anion sensitivity generally exhibited by mammalian other mammalian isozymes has far-reaching implications for homologs. For example, the K for chloride inhibition of bicar- i isozyme-selective drug design. bonate dehydration by hCA IV at pH 5.5 is 36 mM (47), a value lower than the physiological concentration of chloride. In con- We thank Prof. W. S. Sly for kindly providing the affinity-purified mCA trast, mCA XIV exhibited at pH 5.6 an IC of Ϸ250 mM for 50 XIV, Prof. S. J. D. Karlish for helpful advice and ideas concerning kidney chloride inhibition (data not shown), a level significantly higher physiology, and Prof. B. Honig for critical suggestions concerning than that reported for hCA IV, albeit lower than that of the algal electrostatic calculations. The structure was determined in collaboration CAs (18). Thus, CA XIV should remain active under conditions with the Israel Structural Proteomics Center. This work was supported that limit the activity of CA IV. The specific significance of CA by the Israel Ministry of Science and Technology, European Commission IV to bicarbonate absorption in the proximal convoluted tubule Structural Proteomics Project Grant QLG2-CT-2002-00988, and the still remains to be elucidated. Divadol Foundation. A.Z. was supported by Nature Beta Technologies In conclusion, our studies shed light on the molecular adap- (Eilat, Israel) and Nikken-Sohonsha Corp. (Gifu, Japan). J.L.S. is the tations of ␣-type carbonic anhydrases that radically enhance Morton and Gladys Pickman Professor of Structural Biology.

1. Madern, D., Ebel, C. & Zaccai, G. (2000) Extremophiles 4, 91–98. 24. Otwinowski, Z. & Minor, W. (1997) Methods Enzymol. 276, 307–326. 2. Mevarech, M., Frolow, F. & Gloss, L. (2000) Biophys. Chem. 86, 155–164. 25. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., 3. Dym, O., Mevarech, M. & Sussman, J. L. (1994) Science 267, 1344–1346. Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., 4. Frolow, F., Harel, M., Sussman, J. L., Mevarech, M. & Shoham, M. (1996) Nat. et al. (1998) Acta Crystallogr. D 54, 905–921. Struct. Biol. 3, 452–458. 26. McRee, D. E. (1999) J. Struct. Biol. 125, 156–165. 5. Bieger, B., Essen, L. O. & Oesterhelt, D. (2003) Structure (London) 11, 27. Kabsch, W. & Sander, C. (1983) Biopolymers 22, 2577–2637. 375–385. 28. Kleywegt, G. J. & Jones, T. A. (1994) ESF͞CCP4 Newsletter 31, 9–14. 6. Britton, K., Stillman, T., Yip, K., Forterre, F., Engel, P. & Rice, D. (1998) 29. Guex, N. & Peitsch, M. C. (1997) Electrophoresis 18, 2714–2723. J. Biol. Chem. 273, 9023–9030. 30. Nicholls, A., Sharp, K. A. & Honig, B. (1991) Proteins 11, 281–296. 7. Ebel, C., Costenaro, L., Pascu, M., Faou, P., Kernel, B., Proust-De Martin, F. 31. Honig, B. & Nicholls, A. (1995) Science 268, 1144–1149. & Zaccai, G. (2002) Biochemistry 41, 13234–13244. 32. Fisher, M., Pick, U. & Zamir, A. (1994) Plant Physiol. 106, 1359–1365. 8. Costenaro, L., Zaccai, G. & Ebel, C. (2002) Biochemistry 41, 13245–13252. 271, 9. Richard, S. B., Madern, D., Garcin, E. & Zaccai, G. (2000) Biochemistry 39, 33. Fisher, M., Gokhman, I., Pick, U. & Zamir, A. (1996) J. Biol. Chem. 992–1000. 17718–17723. 10. Irimia, A., Ebel, C., Madern, D., Richard, S. B., Cosenza, L. W., Zaccai, G., 34. Stams, T., Nair, S. K., Okuyama, T., Waheed, A., Sly, W. S. & Christianson, Vellieux, F. M., Costenaro, L., Pascu, M., Faou, P., et al. (2003) J. Mol. Biol. D. W. (1996) Proc. Natl. Acad. Sci. USA 93, 13589–13594. 326, 859–873. 35. Whittington, D. A., Waheed, A., Ulmasov, B., Shah, G. N., Grubb, J. H., Sly, 11. Deutch, C. E. (2002) Lett. Appl. Microbiol. 35, 78–84. W. S. & Christianson, D. W. (2001) Proc. Natl. Acad. Sci. USA 98, 9545–9550. 12. Polosina, Y. Y., Zamyatkin, D. F., Kostyukova, A. S., Filimonov, V. V. & 36. Pineda, A. O., Carrell, C. J., Bush, L. A., Prasad, S., Caccia, S., Chen, Z. W., Fedorov, O. V. (2002) Extremophiles 6, 135–142. Mathews, F. S. & Di Cera, E. (2004) J. Biol. Chem. 279, 31842–31853. 13. Wejse, P. L., Ingvorsen, K. & Mortensen, K. K. (2003) Extremophiles 7, 37. Elcock, A. H. & McCammon, J. A. (1998) J. Mol. Biol. 280, 731–748. 423–431. 38. Hakansson, K. (1994) Acta Crystallogr. D 50, 101–104. 14. Yoshimune, K., Yamashita, R., Masuo, N., Wakayama, M. & Moriguchi, M. 39. Chakravarty, S. & Varadarajan, R. (2002) Biochemistry 41, 8152–8161. (2004) Extremophiles 8, 441–446. 40. Whittington, D. A., Grubb, J. H., Waheed, A., Shah, G. N., Sly, W. S. & 15. Madern, D. & Zaccai, G. (2004) Biochimie 86, 295–303. Christianson, D. W. (2004) J. Biol. Chem. 279, 7223–7228. 16. Ahmad, A., Akhtar, M. S. & Bhakuni, V. (2001) Biochemistry 40, 1945–1955. 41. Getzoff, E. D., Tainer, J. A., Weiner, P. K., Kollman, P. A., Richardson, J. S. 17. Avron, M. (1986) Trends Biochem. Sci. 11, 5–6. & Richardson, D. C. (1983) Nature 306, 287–290. 18. Bageshwar, U. K., Premkumar, L., Gokhman, I., Savchenko, T., Sussman, J. L. 42. Ripoll, D. R., Faerman, C. H., Axelsen, P. H., Silman, I. & Sussman, J. L. (1993) & Zamir, A. (2004) Protein Eng. Des. Sel. 17, 191–200. Proc. Natl. Acad. Sci. USA 90, 5128–5132. 19. Premkumar, L., Greenblatt, H. M., Bageshwar, U. K., Savchenko, T., Gokh- 43. Stroppolo, M. E., Falconi, M., Caccuri, A. M. & Desideri, A. (2001) Cell. Mol. man, I., Zamir, A. & Sussman, J. L. (2003) Acta Crystallogr. D 59, 1084–1086. Life Sci. 58, 1451–1460. 20. Lindskog, S. & Silverman, D. N. (2000) in The Carbonic Anhyrases: New Horizons, eds. Chegwidden, W. R., Carter, N. D. & Edwards, Y. H. 44. Jackson, S. E. & Fersht, A. R. (1993) Biochemistry 32, 13909–13916. (Brikhauser, Basel), pp. 175–195. 45. Parkkila, S., Parkkila, A. K., Rajaniemi, H., Shah, G. N., Grubb, J. H., Waheed, 21. Liljas, A., Hakansson, K., Jonsson, B. H. & Xue, Y. (1994) Eur. J. Biochem. 219, A. & Sly, W. S. (2001) Proc. Natl. Acad. Sci. USA 98, 1918–1923. 1–10. 46. Kaunisto, K., Parkkila, S., Rajaniemi, H., Waheed, A., Grubb, J. & Sly, W. S. 22. Jonsson, B. M., Hakansson, K. & Liljas, A. (1993) FEBS Lett. 322, 186–190. (2002) Kidney Int. 61, 2111–2118. 23. Premkumar, L., Bageshwar, U. K., Gokhman, I., Zamir, A. & Sussman, J. L. 47. Baird, T. T., Waheed, A., Okuyama, T., Sly, W. S. & Fierke, C. A. (1997) (2003) Protein Expression Purif. 28, 151–157. Biochemistry 36, 2669–2678.

7498 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0502829102 Premkumar et al. Downloaded by guest on October 1, 2021